Molecular Spintronics using S-Electron...

2008.12.18 TU-Dresden

Molecular Spintronics using -Electron Molecules

Masashi ShiraishiMasashi Shiraishi

1. Osaka University, Japan. 2. JST-PRESTO, Japan

Co-workers in this study

Osaka University

Prof. Yoshishige SuzukiProf. Teruya ShinjoDr. Ryo Nouchi (⇒Tohoku Univ.)Dr. Takayuki NozakiDr. Masaki Mizuguchi (⇒Tohoku Univ.)S. Miwa, H. Kusai, S. Tanabe, M. Ohishi, D. Hatanaka

Experimental Assistance

Tohoku Univ. Univ. of TokyoProf. Yoshihiro Iwasa Prof. Masashi TakigawaProf. Taishi Takenobu Dr. Makoto YoshidaDr. Hidekazu Shimotani

Financial Aid

NEDO, Asahi Glass Foundation, 21st COE Program

Research direction of our Osaka Univ. Group

Molecular Spintronics Molecular Electronics

GrapheneFullerenesCarbon nanotubesOrganic Molecules (Rubrene, Alq3, etc)

Materials

Research Field Research Field

Molecular Science Gp. (Shiraishi’s Group) belonging Prof. Y. Suzuki’s Lab.

CNT-FETsGraphene spintronicsOrganic spintronics…..

Contents of my talk

Introduction : Basic concepts of spintronicsPrevious serious problems in molecular spintronics

Part I : Molecular spintronics using molecular nano-compositesAppearance of magnetoresistance (MR) effectsBackground physicsWhat was missing?

Part II : Molecular spintronics using grapheneSpin injection at room temperature“Local” and “Non-local” MR effectsRemaining problems in molecular spintronics

Background of this study

What is expected to molecular spintronicsWhat is expected to molecular spintronics？？

Molecules ⇒ Carbon, HydrogenA small spin-orbit interaction ( ~ Z4)Long spin relaxation timeLong spin coherent length ?（If high mobility…）

Applications :for instance,

quantum computation devicesmolecular spin transistors

Molecules ⇒ SemiconductorGate-induced (spin) current control

EImp

pIm

Spin-orbit interaction ~A relativistic effect~

Dirac Equation (4×4 matrix)

Positive energy solutions (2-component)

Negative energy solutions (2-component)

m : mass term, : Pauli’s spin matrix, p : momentum

Introduction of an electro-magnetic field (to positive energy solution)

.])2(22

[

]22

)([

2

2

BsLmce

mp

eBmce

meAp

tih

　　　

h

g-value = 2

mpi

i

3

1

.10

01,

00

　i

ii 4×4 matrix

Spin-orbit interaction ~A relativistic effect~

Diagonalization of the Dirac equation by Unitary transformation(Excluding the mixing of positive and negative solutions)

⇒ Foldy=Wouthuysen transformation

divEmepE

merotE

mie

Bmee

mp

meApmH

222

2

42

848

)(2

}82

)({

　　　　

Mass velocity term Electrostatic dipole

Magnetic dipole

Spin-orbit interaction terms Darwin term(Zitterbewegung)

To spherically symmetric potential, the 1st term = 0.The 2nd term is double of that in Classical mechanics (Thomas precession).S-O Hamiltonian is

)(4 3 L

mrZe

H B h ⇒ Z4

Alq3

Co

LSMO

Difficulty in obtaining a clear interface & implementing various characterizations.

Underlying physics “?”

Spins were injected????

Co

Alq3

Correspondence between MR effects and magnetization of FM

Spurious signals (Large Rc)X

In-plane structures Sandwiched structures

How can we reproduce them ?

Molecular spintronics chronicle (1)Serious problems in molecular spintronics

Molecular spintronics chronicle (1)FM FM

Molecule

① Resistance hysteresis & Magnetization Process

②

Local Hall effect

Magnetic domain wall

③

⑤ Reliability of a “local” scheme “?”

④ Large noise (Rc)⇒ Spurious signal

B

R : Small R : large

Serious problems in molecular spintronics

B

R

B

Anisotropic MR effect

Molecular spintronics chronicle (1)How to solve or bypass the problems?

①

②

③

MR curves & Magnetization processare characterized in the same sample.

TMR oscillation

Non-local method(excluding spurioussignals)

PART I : Molecular nano-composite spin devices

Remaining obstacles & our strategy

1. A lack of detailed analyses of magnetization processes（Device structures were not suitable for various characterizations.）

2. Excess contact resistance ( ⇒spurious signals )

3. Uncontrolled interface formations between FM/NM

should be solved for further progressand obtaining reliable results !!

Introducing a novel system, C60-Co nano-composites⇒ bypassing the interface problems

inducing reliable results by various characterizationsH. Zare-Korsaraki and H. Micklitz, 2004

In this study…

Fabrication processes and sample structures

1. Using photo1. Using photo--lithographylithography2. Evaporating Cr and Au electrodes2. Evaporating Cr and Au electrodes3. Lift3. Lift--offoff4. 4. CoevaporationCoevaporation of Cof C6060--Co Co ((1010--66~10~10--7 7 TorrTorr))5. Capping layers 5. Capping layers ((CC6060++SiOSiO))7. Coating by ZEP7. Coating by ZEP--520A520A8. Annealing at 180 8. Annealing at 180 ℃℃, 30 min in vacuum, 30 min in vacuum9. Magnetoresistance (MR) measurements 9. Magnetoresistance (MR) measurements

C60evaporation systems C60 powderCo

4.9 eV

C60

6.2 eV(Shinohara, CPL)HOMO

V.L.

LUMO

~ 1.5 eV(S. Saito PRL)

4.7 eV

Small gap between W.F. of Co and LUMOcomparing with a Alq3 case.

The MR curve coincides with the magnetization of the Co.

Magnetization of the Co induces the observed MR effect.= Reliable study !

3 nm 2 nm

Introduction of C60-Co nano-composites

S. Miwa, M.S. et al. JJAP 45, L717 (2006).S. Miwa, M.S. et al. PRB76, 214414 (2007).

Structural analyses of the C60-Co nano-composite

1. The Co size：2-3 nm (Fig.1,XRD：2 nm）

2. Distance ~ 1.5-2.2 nm (From XRD)3. No percolation of the Co（Fig.1）

4. Spin dependent transport in C60（Fig.2）

5. No obvious damage to C60（Fig. 3）

Fig. 1 Fig. 2

Fig. 3

1426

14601469

1573

3 nm 2 nm

Temperature dependenceof the MR ratio(= 290 K)

An observation of the MR effect at room temperature

The first observation of an MR effect at RT induced by spin-dependent transport in molecular spin devices !!

Recently,….the MR ratio increased up to 300-500% at 4 K.

A role of molecules ?

T-1/2 dep. (Hopping transport) ⇒ Coulomb interaction between conducting electrons

Co-Al-O granular filmS. Mitani et al., PRL 1998.

In Co-Al-O system,

Molecules behave as a tunnel barrier ?

Coulomb energy

When T=6.5 K, thenEc=1.48 meV

Number of junctions :300-400

Application for field effect transistors (FETs)S.C. FET = ~40 cm2 V-1 s-1

V. C. Sundar et al., Science 312, 1644 (2004).J. Takeya et al., Appl. Phys. Lett. in press.

Thin film FETS. Seo, et al., Appl. Phys. Lett. 88, 232114 (2006).

Rubrene : C42H28

Rubrene~ 2 nm

Replacement of molecular matrices

Alq3

An electron transport layer of O-LEDTang et al., Appl. Phys. Lett. 51, 913 (1987).An MR effect in sandwiched devicesXiong et al., Nature 2003An intrinsic MR effect （Wohlgenannt et al.)

① Fabrication of NM electrodes② rubrene/Co co-evap.(10-7~10-6 Torr) rubrene : ~ 0.3 Å /s , 1000 Å

Co : ~ 0.1 Å /s , 350 Å③ Capping layers [SiO (~ 500 nm) + resist]

Volume contentsrubrene：Co = 3 : 1

H

GlassAu/Cr (50 nm)

gap 10 m

4 mm

4 mm 4 mm

Au/Cr (50 nm)

Co

rubrene

SiOrubrene/Co

135 nm

Resist

A device structure using rubrene-Co nano-composite

7

6

5

4

104

m)

-40 -20 0 20 40H (kOe)

-1.0

-0.8

-0.6

-0.4

-0.2

0.0-(M

/Ms ) 2H

: resistivityMs : sat. mag.

4.2 KVbais = 2 V

MR ratio =min

max - min = 78%

-(M/MS)2

Electrode

Corubrene

7.06.5

10

m

)

-1.0 0.0 1.0H (kOe)

-0.3-0.2-0.10.0

-(M/M

s )

A large MR effect in rubrene-Co nano-composite

Electrode

H. Kusai, M. Shiraishi et al., Chem. Phys. Lett. 448, 106 (2007).

MR ratio =

Co spin polarization: P = 34% → MR ratio = 12%

C60-Co: MR ratio = 300%Rubrene: MR ratio = 80%

1. Enhancement of the MR ratio by Coulomb blockade ? (S. Mitani et al., PRL81, 2799 (1998).)

2. Enhancement of spin polarization of the Co at the interface ?(C-K. Yang et al., PRL203 (2003).)

Larger than the theoretical predicted value

Possible reasons

P2

The origins of the large MR ratio

1. Yes. Coulomb blockade and co-tunneling are important.2. Yes. (M. Shiraishi et al., APL 93, 53103 (2008). 59Co spin echo)

Importance of surface & interface characterizations !!

Why Coulomb blockade is important?

A spin flip process is needed

Low conductivity !!

Parallel alignment Anti-parallel alignment

1

0

-1

Cur

rent

[mA

]

-2 -1 0 1 2Bias voltage [mV]

3.2 K 20 K 40 K 80 K

Clear Coulomb blockade

10-9

10-7

10-5

10-3

G0

[1/o

hm]

0.120.100.080.060.040.021/T [1/K]

TkUG

B

exp

U : 12 meVrCo : 1.4 nm

rCo ：0.8-2.3 nm (by SQUID)

V=10 mV

dAu-Au=400-500 nm

Zero bias conductance :

NOT breakdown !Spin (Coulomb) blockade !!1.1 V

1.00.50.0

-0.5-1.0C

urre

nt [m

A]

-2 -1 0 1 2Bias voltage [V]

60

40

20

0

MR

ratio [%]

Current MR ratio

Correspondence between CB and MR

Due to Spin blockade

??? Co-tunneling?Two-step enhancement !!

T=3.2 K

Bias voltage dependence of MR ratio (80 K)

■ The I-V characteristic is linear, thus no CB effect.

■ The MR ratio monotonously decreases with increasing the bias voltage

80 K

-2

-1

0

1

2C

urre

nt [m

A]


0.32

0.28

0.24

0.20

MR

ratio [%] Current

MR ratio

15 K

-2

-1

0

1

2C

urre

nt [m

A]


6

4

2

0

MR

ratio [%] Current

MR ratio

Bias voltage dependence of MR ratio (15 K)

■ Constant MR ratio in the CB region → Second-order co-tunneling

■ Clear appearance of the MR enhancement by the CB effect

S. Takahashi and S. Maekawa, phys. Rev. Lett. 80, 1758 (1998).

e-e-

Bias voltage dependence of MR ratio (3.2 K)

■ Clearer appearance of the enhancement by the CB effect at the threshold

■Appearance of an additional MR enhancement in the CB region

3.2 K


-2

-1

0

1

2C

urre

nt [m

A] 50

403020100

MR

ratio [%]

Current MR ratio

Log-log plots of I-V characteristics

■ Agreement with the transport property of Co-tunneling (I ∝ V2N-1 , N :the number of junctions)

■ N decreases with increasing temperature.

Dominant transport : High-order Co-tunnelingT. B. Tran et al., Phys. Rev. Lett. 95. 076806 (2005)

10-7

10-5

10-3

10-1

Cur

rent

[mA

]

2 3 4 5 6 7 8 91

2Bias voltage [V]

80 K 40 K 20 K15 K12 K10 K8 K6 K4.2 K3.2 K I V9

I V3

Conclusion 1 (CB & CT)

1. Clear Coulomb blockade (CB) and co-tunneling (CT) were observed.

2. Two-step enhancement of the MR ratio was observed.

3. The CB & the CT contributed to the enhancement of the MR ratio.

D. Hatanaka, M. Shiraishi et al., to be submitted.

PEN(polyethylene naphthalate) substrate(50 m thickness)

85 mm

90 mm

kapton film mask

5 mm10 mm PEN substrate

glass cap

Sample layer

tetrafluoroethylene film

・Co(~350 nm, 0.1 Å/s) + glass・rubrene(~1050 nm) + glass・rubrene(~1050 nm)-Co(~350 nm) + glass・rubrene(~700 nm)-Co(~350 nm) + glass

PEN substrate：10 mm x 5 mm x 48 pieces

Measured Samples

Spin polarization characterization (NMR spin echo)

Spin polarization characterization (NMR spin echo)

59Co spin echo

Co nano-particles in the nano-composites(Diameter：0.8-2.3 nm)

No signal from rubrene

Temperature： 2 KFrequency：100-320 MHzMagnetic field：0-9 T

hcp-Co

Spin echo signals (Co thin film & rubrene-Co)

Enhancement of spin polarizationat the Co surface.

~ +10 %

hcp-Co

Spin echo signals (Magnetic field dependence)

Completely different behavior in comparison with Co thin film !!

Signals become weaker !!

An interpretation

Important message）Unique behavior under a magnetic field application

=A possible model=①B=0 T

A hyperfine field is fully observed.

②B > 1 TParallel to the applied field,

but random to the crystal axis. ⇒ anisotropy of the hyperfine field

+hcp-Co induced a quadrupole effect

Conclusion 2

1. Enhancement of spin polarization was observed by NMR.

2. The enhancement was estimated to be ~10%.

3. It is not enough to explain all of the enhancement of the MR ratio.

4. The spin structure of molecular spin devices was firstly clarified.

M. Shiraishi et al., Applied Physics Letters 93, 53103 (2008).

1. The first observation of a spin-dependent MR effect at RT in molecular spin devices.

2. Large MR ratio ( 80%) was observed in rubrene-Co nano-composites.Recently, 300-500%@4 K in a C60-Co system.

3. MR ratio of 12% in Alq3-Co nano-composites.(S. Tanabe, M. Shiraishi et al., APL 91, 63123 (2007).)

4. Background physics becomes clear.

Conclusive remarks in Part I

However, spin injection was not achieved !!・ Molecules are barriers (TMR!)・ Conductance mismatch

Sakai, Takanashi, et al., APL 2006. (C60, 400%@4 K, TMR)Moodera et al., PRL 2007. (Alq3, Rubrene : TMR) Bader et al., PRB 2008. (Alq3 : TMR)

FM/NM contactSpin-injection

μ↑

Ferromagneticmetal

non-magneticmaterial

J

μ↓

z

μ

μ↑-μ↓

z

Spin accumulation

nmlFM 5～ nmlNM 100 several

FM

NM

rrJJ

JJ

1

Spin-polarization of the injected current

:Spin-asymmetry:interface effectiveresistance

/4/11

FMFM lr2/1

NMNMNM lr

FMNM rrWe cannot inject spin for the case;

A. A. FertFert et al., PRB 2001.et al., PRB 2001.

Molecular nano-composite spin devices

PART II : Spin injection into graphene at room temperature

How to inject spins into molecules ?

The answer is….

Graphene

1HOPG

HOPGScotch tape

remove

2

SiO2/++Si

push

3

Thin film Graphene

SiO2/++Si

Fabrication process (Scotch tape peeling)

Graphene (Single- & Multi-layered, Optical microscopic image)

Sample geometry and measurement techniques

Local 2-T & Non-local 4-T DC measurements

Room temperatureNo tunneling barrier

RCo1-Au1 ~ 110

(RCo-wire ~ 50 )

RCo1-Co2 ~ 205

The basic concept

F.J. Jedema et al., Nature 416, 713 (2002).

spin injector

spin detector

An electric current : anisotropicA spin current : isotropic

Al nanowire: non magnetic

A proof of spin injection into the Al nanowireusing difference of electrochemical potential of spins

A concept of “non-local” measurements

Observation of spin injection signals

Jpn. J. Appl. Phys 46, L605 (2007). (Express Letter)

Clear proof of“spin injection” into graphene !!

0.716

0.717

0.718

0.719

0.720

0.721

-40 -20 0 20 40

V non-

loca

l/I (V

/A)

B (mT)

185.14185.15185.16185.17185.18185.19185.20185.21

R loca

l ()

Local 2-T

Non-local 4-T

Magnetic domain wall

Electric current

×AMR effect

Local 2-T : AMR-induced signals Non-local 4-T : Spin injection signals

⇒ Detection of spin current in graphene at room temperature

Spintronics using graphene2007.06.22 Our paper was published !!Cond-mat) 0706.14511451Multi layer graphene (Graphene Thin Film, GTF) W/O tunneling barrier

2007.07.15Groningen Univ., Nature 448, 571 (2007).Cond-mat) 0706.19481948Single layer graphene WITH tunneling barrier

2007.09.19Univ. of Maryland, APL 91, 123105 (2007).Cond-mat) 0706.15971597Single layer graphene W/O tunneling barrier

The other repotsby M. Nishioka et al. (in APL)by R.K. Kawakami et al. (in PRB) and more…

Inferiority of molecular spintronics (1)

MgO-based TMR : ~1 V (RT)S. Yuasa et al., Nature Mat. 2004.

Vhalf : MR ratio becomes a half of max.The decrease : spin scattering = loss of P

(magnon/phonon…)

SWNT spin valve : ~10 mV (4 K)MR ratio ~ 2 %H.T. Man et al., PRB 2006.

graphene spin valve : ~ 5 mV (2 K)MR ratio ~ 0.4% M. Nishioka et al., APL 2007.

Poor bias voltage dependence

No accordance between “Local” & “Non-local” results

“local” R~700 “non-local” R~20

N. Tombros, B.J. van Wees et al., PRB 2006.

“…90% of local signalsare not due to spinaccumulation….”

The missing part inmolecular spintronics(Underlying physics cannotbe discussed.)


T=4.2 K

２×（Non-local R）＝(Local R)

F.J. Jedema et al., PRB 2003.

Tombros, van Wees et al., Nature 2007.

“local” R > 100

“non-local” R~12

Jedema, van Wees et al., PRB 2003.

“non-local” R = 2 m

“local” R = 4 m

Molecule Metal


T=5 K

Next milestones

Solving the two important problems in molecular spintronics;

1. Poor bias voltage dependence of spin polarization2. Missing accordance between “local” & “non-local” results

by using our graphene-based spin valves.

The milestone ;

Constructing a steadfast basis of molecular spintronics

Local magnetoresistance at room temperature

M. Shiraishi et al., cond-mat 0810.4592.

Spin precession and spin coherence in “non-local”

ComparisonComparison

Diffusion constant, D (10-2m2/s)

Spin flip length, m

Spin coherent time, (ps)

Spin polarization, P

120 K RT RT (Dirac point)

0.8 2.1 1.3(P)~2.1(AP)

1.1 1.6 1.3(P)~1.5(AP)

150 120 125(P)~100(AP)

0.16 0.09 0.1

Our study (GTF)Our study (GTF) Tombros et al. (Single layer)

The previous report from our group (2)

Molecular spintronics chronicle (1)Injection current dependence of output signals

“Non-local” “Local”

Injected current dependence of output voltages is linear.

P is constant !! = Robustness of spin polarization !!

),exp(2

2

sfinject

GTF

sflocalnon

LIA

PV

Robustness of spin polarization (1)

Comparison with that in MgO-TMRP was constant up to 0.5 V, and was still 81% at 1.2 V


M. Shiraishi et al., cond-mat 0810.4592.

Molecular spintronics chronicle (1)

Unprecedented robustness…?Suppression of Magnon/Phonon excitation?

２×（Non-local R）＝(Local R)


Ideal interface formationIdeal interface formation

More important message isMore important message is……..““Theory and experiment firstly Theory and experiment firstly exhibit the good accordanceexhibit the good accordancein molecular spintronics.in molecular spintronics.””

Room temperature !!

We have succeeded in constructing a steadfast basis of molecular spintronics.

Conclusive remarks in Part II (graphene spintronics)

1. Spin injection and spin current detection in graphene(by non-local measurements)

2. Magnetoresistance effect at room temperature(by non-local & local measurements)

3. Estimation of spin transport properties(by Hanle-type spin precession)

1. Reliable results are provided.2. Unprecedented robustness of spin polarization is found.3. A theory and an experiment coordinate well

in molecular spintronics.

Thank you very much for your kind attention !!

0.993

0.994

0.995

0.996

0.997

0.998

0.999

1

-400 -300 -200 -100 0 100 200 300 400

Magntic Field (Oe)

"Half"

0.22

0.221

0.222

0.223

0.224

0.225

0.226

0.227

0.228

-400 -300 -200 -100 0 100 200 300 400

Magnetic Field (Oe)

"Cross"

Spin injection signals in graphene spin valves (1)

5 V 5 V

800 A 800 A

Spin injection signals in graphene spin valves (2)

Non-local signals in “cross” and “half” alignments.

If both are different…. Ohmic contact

If both coincide…. Schottky contact(T. Kimura et al., JMMM 286, 88 (2005).)

Schottky ? Conductance mismatch ?

Wire resistance : 100 Graphene resistance : 5 Device resistance :200-300

Contact resistance ? : 100

Unintentional contact resistance exists ? Is it NECESSARY or NUISANCE ?

5.0245.0225.0205.0185.016 1

0-4

m)

-40 0 40H (kOe)

-1.0

-0.5

0.0 -(M/M

s ) 2V = 100 mV

MR比 = ~ 0.1%

Observation of the MR effect at RT.

Drastic decrease of the MR ratio

Sat. mag. does not induce the decrease.

4.2 K → RTSat. mag. decreased only 30%

RT

-(M/MS)2

0.1

1

10

100M

R (%

)

2 410

2 4100

2 4

T (K)

Temperature dependence of the MR ratio

80

60

40

20

0

Ms (

emu/

g C

o)

3002001000T (K)

MR ratio4.2 K (68%) → RT (0.1%)Sat. mag.4.2 K (74 emu/g Co)

→ RT (54 emu/g Co)

Temperature dependence of the saturation magnetization

Molecular Spintronics using S-Electron...

Documents

Transcript of Molecular Spintronics using S-Electron...